This repository contains the fast-agent source files used in the pulse-mcp demo.
To run the examples, install the uv package manager then
uv pip install fast-agent-mcp
Update the fastagent.secrets.yaml with your API Keys. Run agents with uv run agent.py.
These files are created with the fast-agent setup and fast-agent bootstrap workflow commands.
A demonstration of MCP Prompts and Resources showing in context learning and a PDF Resource being returned within a prompt. fast-agent's built in prompt-server is used to deliver the prompts to the agent.
An example of a sampling being used to generate a dynamic resource (resource templating).
Example of vision sampling via mcp-webcam, as well as the use of vision to make agent routing decisions. MCP Webcam, Home Assistant and Hugging Face Spaces Servers are used in the demonstration.